Time Domain Structural Fatigue Analysis of Floating Offshore Platforms: A Response Based Technique

Abstract

Abstract This paper presents a response-based, time-domain structural fatigue analysis of a floating offshore platform. The conventional technique for structural fatigue assessments of offshore platforms uses a linear, frequency-domain analysis based on the spectral method. Although this conventional method is computationally efficient, there is a room for improving accuracy and reducing uncertainties because it cannot accurately address non-linear loadings on the offshore platform. Such non-linear loads arise from the wave, wind, and current as well as from the riser and mooring systems; these non-linearities necessitate large factors of safety that lead to conservative design and frequent inspection. As an extension of previous work (Kyoung et al.[12]), this study presents the development of a time-domain, structural fatigue analysis that explicitly addresses non-linear loading on the platform. The external load time-histories are directly mapped onto the structure at every time interval to create a stress-based response with the varying environment. In each time step, the load mapping accurately captures the phase relationship between the external loading and hull inertial response. Therefore, present method reduces uncertainties in the fatigue damage computation and overcomes the assumptions of spectral method. Present load component-based approach is applied onto a finite element structural model, which provides unit structural response at locations of interest. Time history of structural response is obtained by synthesizing the obtained unit stress-based structural response with environmental loading and platform motion response. Fatigue damage can be computed from the obtained time series of structural response using rain-flow counting. As an application, a conventional semisubmersible platform is used to evaluate structural fatigue damage for a given wave scatter diagram. A comparison between results from this response-based time-domain approach and the conventional spectral method is presented.